Synopsis: A program GALILEO was built that detected and repaired conflicts between theories of Physics and experimental results. It used Ontology Repair Plans (ORPs) to formalise common patterns of fault detection and repair in Physics and other domains.
Hypotheses: Automated ontology evolution via ORPs is computationally feasible and can account for the kinds of ontology evolution that are observed in human problem solving in the Physics domain. We show that desirable properties, e.g., coverage, efficiency, maintainability, high quality of the repairs, can be achieved.
Evaluation: The presentation, application and evaluation of three of GALILEO’s Ontology Repair Plans provided empirical evidence of the value of our approach to ontology evolution. The ORP Where’s My Stuff? was applied to the discovery of latent heat and to the postulation of dark matter; Inconstancy was applied to Modified Newtonian Dynamics in the study of galaxies and to the observations that proved the speed of light to be finite; Unite was applied to the identification of the Morning and Evening Stars and to the assessment of the shape of the Earth.